sort file list lexicographically in detect() for deterministic graph output

os.walk order is non-deterministic across runs (filesystem b-tree order
shifts with cache/mount state), causing first-writer-wins decisions in
cross-file resolution to produce different node IDs each run; sorting
all_files and per-FileType lists stabilises graph.json without changing
any parent-child or connectivity semantics (closes #1090)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

graphify/detect.py

@@ -930,6 +930,8 @@ def detect(root: Path, *, follow_symlinks: bool | None = None, google_workspace:
                     seen.add(p)
                     all_files.append(p)
 
+    all_files.sort(key=lambda p: str(p))
+
     converted_dir = root / "graphify-out" / "converted"
 
     for p in all_files:
@@ -983,6 +985,9 @@ def detect(root: Path, *, follow_symlinks: bool | None = None, google_workspace:
             if ftype != FileType.VIDEO:
                 total_words += count_words(p)
 
+    for ftype in files:
+        files[ftype].sort()
+
     total_files = sum(len(v) for v in files.values())
     needs_graph = total_words >= CORPUS_WARN_THRESHOLD